Pulmonary tuberculosis screening in anti-retroviral treated adults living with HIV in Kenya

Abstract

Background People living with HIV (PLHIV) who reside in high tuberculosis burden settings remain at risk for tuberculosis disease despite treatment with anti-retroviral therapy and isoniazid preventive therapy (IPT). The performance of the World Health Organization (WHO) symptom screen for tuberculosis in PLHIV receiving anti-retroviral therapy is sub-optimal and alternative screening strategies are needed. Methods We enrolled HIV-positive adults into a prospective study in western Kenya. Individuals who were IPT-naïve or had completed IPT > 6 months prior to enrollment were eligible. We evaluated tuberculosis prevalence overall and by IPT status. We assessed the accuracy of the WHO symptom screen, GeneXpert MTB/RIF (Xpert), and candidate biomarkers including C-reactive protein (CRP), hemoglobin, erythrocyte sedimentation rate (ESR), and monocyte-to-lymphocyte ratio for identifying pulmonary tuberculosis. Some participants were evaluated at 6 months post-enrollment for tuberculosis. Results The study included 383 PLHIV, of whom > 99% were on antiretrovirals and 88% had received IPT, completed a median of 1.1 years (IQR 0.8–1.55) prior to enrollment. The prevalence of pulmonary tuberculosis at enrollment was 1.3% (n = 5, 95% CI 0.4–3.0%): 4.3% (0.5–14.5%) among IPT-naïve and 0.9% (0.2–2.6%) among IPT-treated participants. The sensitivity of the WHO symptom screen was 0% (0–52%) and specificity 87% (83–90%). Xpert and candidate biomarkers had poor to moderate sensitivity; the most accurate biomarker was CRP ≥ 3.3 mg/L (sensitivity 80% (28–100) and specificity 72% (67–77)). Six months after enrollment, the incidence rate of pulmonary tuberculosis following IPT completion was 0.84 per 100 person-years (95% CI, 0.31–2.23). Conclusions In Kenyan PLHIV treated with IPT, tuberculosis prevalence was low at a median of 1.4 years after IPT completion. WHO symptoms screening, Xpert, and candidate biomarkers were insensitive for identifying pulmonary tuberculosis in antiretroviral-treated PLHIV.

Tuberculosis screening among ambulatory people living with HIV: a systematic review and individual participant data meta-analysis

“…Individual participant data were provided for 22 studies (including one study 13 that was eligible only for sensitivity analyses). 6 , 10 , 11 , 12 , 34 , 35 , 36 , 37 , 38 , 39 , 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 Individual participant data were not provided for three studies. 51 , 52 , 53 Hence, we obtained individual participant data for 15 666 (92%) of 17 024 participants identified.…”

“…Only one study (n=381) among outpatients on ART assessed C-reactive protein (≥10 mg/L), for which there was a similar sensitivity and specificity compared with W4SS alone ( appendix p 18 ). 49 …”

“…Only one study (n=381) among outpatients on ART assessed C-reactive protein (≥10 mg/L), for which there was a similar sensitivity and specificity compared with W4SS alone (appendix p 18). 49 We did sensitivity analyses using two alternative reference standards: culture or Xpert, and Xpert alone (appendix pp [47][48][49][50][51][52]. Results were largely similar to the main analyses, although sensitivities were slightly higher for the reference standard of Xpert alone than for the main reference standard of culture.…”

Screening performance of C-reactive protein for active pulmonary tuberculosis in HIV-positive patients: A systematic review with a meta-analysis

“…Thirteen studies were included in the meta-analysis that evaluated CRP for PTB among HIV outpatients ( 14 , 16 , 18 , 19 , 41 , 46 , 47 , 49 – 54 ). The studies provided data for 4,355 HIV-positive adults, including 891 (20%) with PTB.…”

“…Gersh et al. assessed 0% sensitivity, 87% specificity, and 99% negative predictive value of the WHO-4-SS algorithm, concomitant with 20% sensitivity, 99% specificity, and 99% negative predictive value of the Xpert Gene test for identifying TB in 383 HIV-infected individuals ( 19 ). Therefore, the WHO-4-SS screening tool is considered partially effective in HIV-infected patients ( 17 ).…”